Re: [RDFTM] preliminary draft

Dear Pat,

many thanks for your comments.
I'll fix the document at my best considering them.

>
> If I might make a suggestion, it is usually less work in the long  
> run to work out the formal mappings first, and then make sure that  
> the informal account follows the mathematics. Informal rules made  
> without paying attention to the underlying foundations are almost  
> never correct.
>
>
You're probably right.
Our intention is to go through the formal specification after the f2f  
meeting as we have to decide which formalism to use for this aim.

> A first quick look prompts the following immediate comment.
>
> The translation from RDFS to TM is defined in such a way that it  
> depends on information which may not be available, in particular  
> whether a resource is a member of the class  
> rdftm:InformationResource. What is the intended translation  
> technique to be used when membership in this class is unknown (as  
> it will often be, presumably)?
>
Thank you for pointing out this.
The approach for RDF/TM interoperability is twofold.
On one hand we have unguided translation, on the other hand we have  
guided translation.
This preliminary draft is about "guided" translation. In fact it is  
based on the rdftm: vocabulary which is used for explicitly express  
how the translation has to be performed.
We chose first to deal with guided translation.
This is why we assume to have such information available to those rules.
Probably, this is not clear by the current version, I'll try to do it  
soon.

> A related point, the document gives as an RDFTM 'rule',
>
> "If the resource is not an instance of rdftm:InformationResource,  
> the resource becomes a subject locator..."
>
> How is it possible to determine in RDFS that a resource is NOT an  
> instance of a class? Such information can be expressed in OWL, but  
> not in RDFS. It is not valid to infer non-membership in a class  
> from a failure to prove membership in that class. The translation  
> rule as stated seems therefore to be inapplicable to RDFS-to-TM  
> translation, or ill-defined, or possibly both.

I agree with you. Do you think it might be better to say:
"If the resource is not explicitly an instance of  
rdftm:InformationResource....." pointing out that we are referring to  
the guidance.
To declare that a resource is an instance of  
rdftm:InformationResource is a way to say how to handle the URI of  
that resource in Topic Maps. Not to do it give a the same information  
leading to a different result.


>
> I will send other comments later when I have had time to study the  
> document more carefully, but these points seem central and require  
> immediate attention.
>

Thank you very much for your help.

Valentina




-----------------------------------------------------------
Valentina Presutti

Phd Student of Computer Science
Department of Computer Science
University of Bologna
Mura anteo Zamboni 7
Ph. +39 (0)51 20 94871
Fax. +39 (0)51 20 94510
home page: http://www.cs.unibo.it/~presutti